Google will most likely will win AI race (and maybe AWS too)
Update: I wrote this before Gemini 3 came out. I just never published it. However, now that Gemini 3 is out, it just proves my point even more. Even OpenAI is freaking out and calling “code red” in an internal memo.
Last month I put my entire month’s paycheck into Google stock. It’s because I firmly believe that Google is going to win the AI race. I also think Amazon/AWS will be a close second.
The ChatGPT brand is really closely associated with AI itself. When someone says “just ask AI” or “chat,” they’re most likely referring to ChatGPT. This is because the launch of ChatGPT is what actually started this entire AI wave (specifically I believe ChatGPT 3.0 was the first real model that performed decently well).
However, since then, competition has become fierce:
Anthropic’s Claude models such as Opus 4 are breaking coding benchmarks
Open source models being launched by Chinese labs for a fraction of the cost, like DeepSeek and Qwen, are punching at the same level as the GPT series
Now Google has two extremely impressive models: Gemini 2.5 Pro and 2.5 Flash, which are still state of the art (SOTA) models
From what I see, ChatGPT’s claim to fame is no longer its intelligent models, but the fact that it’s the brand behind AI.
But now Google’s coming for the throne.
Google’s traffic is still 34x ChatGPT’s. They’re the default search engine across billions of iPhones and Androids across the globe. And they offer this service all with zero cost to the customer.
With the additional friction of having to navigate to chatgpt.com instead of it being the default way users search, ChatGPT is already at a disadvantage in terms of accessibility.
Google offers their entire ecosystem free to customers. This is because they’ve built one of the most sophisticated advertising platforms in the world that’s trained on everything about a user: Google Search, Google Photos, Google Drive, Google Maps, YouTube, etc.
Only 5%, or 35 million people, pay for a ChatGPT subscription. ChatGPT is reportedly attempting to implement advertising into its platform, but the success of this is still to be demonstrated. What will they target ads on? ChatGPT conversation history? This is drastically insignificant compared to Google’s targeting technology.
Google has already started introducing AI responses into its search results, right at the top. I’m guessing this is powered by Gemini 2.5 Flash. The results are very impressive. It basically answers any “quick” questions I really need it to. And since it’s also a search engine, it’s able to recommend links and YouTube videos that may be relevant. I find myself only navigating to AI native interfaces like Claude or ChatGPT when I need to do deeper research or have a back and forth conversation. Obviously, for tasks like coding, I’m going directly to Claude or Cursor.
Google needs to just master the actual user experience. They need to somehow integrate the search and chat functionalities seamlessly. I’m thinking maybe after searching, they display the AI response, and under it users can continue the conversation, transitioning into a conversation window perhaps. This will allow them to continue operating as a search engine “enhanced” by AI.
Google can do this all without charging their users. They have the capability to monetize it with ads. They don’t have to worry about luring users to pay $20 a month—which by the way, lots of people aren’t willing to do.
Google is not only at the advantage of being at the fingertips of billions of people already while having an already profitable monetization model in place, but also has its own chip called the Tensor Processing Unit. Its latest chip, called Trillium, has 24x more power than the fastest supercomputer to date. Google has no reliance on third party chip manufacturers like NVIDIA. However, OpenAI solely relies on NVIDIA chips. They’ve signed this insane deal with them, which is only increasing their capex and tightening their margins (which are already in the negatives btw). OpenAI is now in this toxic relationship with NVIDIA, because it can’t even start its own chip manufacturing in-house—otherwise they risk ruining that relationship entirely, which they cannot afford to do as they need those chips to continue their AI model development.
Last thing Google has over OpenAI is cloud infrastructure. Yes, AWS surpasses Google Cloud in my opinion in cloud offerings, but don’t be fooled by Google Cloud. Google’s cloud offerings are eating up market share every year. Their services are resilient and are already operating at a global scale. Google can operate their models on their own infrastructure at their own cost. Unlike OpenAI, which has to make deals with companies like AWS to operate their models.
I don’t know how OpenAI wins this battle against Google. Time will tell I guess.
I notably see AWS positioned to be the second winner in the AI race. Not because I think they’ll be a model provider, but solely because of their cloud infrastructure which they can offer to these different model providers. Anthropic and OpenAI are two notable providers already using AWS for LLM inference. Amazon already owns 15-19% of Anthropic. I believe AWS will win the AI compute-as-a-service race, while Google will just win the model provider race (or both maybe).
So for now, I’ll be loading up on $GOOG. This is not financial advice. I am not a financial advisor. I am Faraaz Khan.
